Google wifi system for isolation?

People have done this with Apple Airports.

MangoMonkey posted:

......... made a big positive improvement to the noise floor.

......... Again, a solid reduction both in noise floor, as well as in some hotness in the music that I've gotten used to, but am glad when I don't hear it. 

How do you measure noise floor MM,  or is it just a feeling/hearing/perception,  if so please can you describe what you are hearing.   I ask because I'm struggling to measure system noise - or at least get consistent results - with an oscilloscope, I see it but don't hear any audible change when what I see is not there.   

What's often referred to as 'noise', is usually low-level non-harmonically related signal components.

Huge posted:

What's often referred to as 'noise', is usually low-level non-harmonically related signal components.

Whoooo   that new avatar threw me,  which one are you ??

Re my question on 'noise',    needless to say I'm playing with ferrets on the SMPS's again,  I've junked those I'd used out of my bits&pieces box  that were mostly left over's from a previous life,  I had assumed them to be for low Hz,  but I'm not sure & you know what it is with OCD doubts,  a scratch that has to be itched.     

Mike-B posted:

Whoooo   that new avatar threw me,  which one are you ??

<snip>

Neither, I'm a member of a different group who sometimes face prejudice and discrimination.  If you come to the "Forum Meet-up" HH is planning to arrange, and you meet me in real life, then you'll see to what I refer (hence also not posting my 'official/legal' name on here).

Mike-B posted:

Huge posted:

What's often referred to as 'noise', is usually low-level non-harmonically related signal components.

<snip>

Re my question on 'noise',    needless to say I'm playing with ferrets on the SMPS's again,  I've junked those I'd used out of my bits&pieces box  that were mostly left over's from a previous life,  I had assumed them to be for low Hz,  but I'm not sure & you know what it is with OCD doubts,  a scratch that has to be itched.     

To what breed of ferrets are you changing?

MangoMonkey posted:

At this point, I often wonder whether I should have stuck to the CD5XS into nDac instead of getting into streaming. Vinyl via the Rega TT is fiddle free in comparison - and has a lower noise floor than the streamer based systems.

And FM radio too, for a nice set.

Vinyl, FM radio...

Lower noise floor?

There's something very wrong with your systems then.

Huge posted: 

To what breed of ferrets are you changing?

I'm changing some split clamps I had plastered around my SMPS's on both 230VAC & DC sides.   They originally came from my work life that we used to damp DC feeds into a data logger & satellite comms system that objected to a 100 KHz DC ripple problem,  & as such I assumed they were one of the low frequency mixes around in those days.  As most all ferrite mixes are ineffective below 1 MHz & I really did not know what the mix was, the doubt was a scratch that I had to itch.     

SMPS switching frequency is typically 25 KHz to 2 MHz, some going to 3 MHz, most common I read on www is 50 kHz & 1 MHz,  however the big problem with ferrite is none of the mix formulas work in the lowest 25 KHz frequencies of the SMPS switching band.     I've opted for Fair-Rite 75 as its the lowest frequency ferrite mix in split clamp applications.  it has a maximum impedance peak around 2 MHz where it's ideal for the SMPS max switching Hz point. Although its progressively less effective as it goes down to its lower limit of 100 KHz,  they will impart some suppression.  I have the SMPS DC wires wound around the ferrite cores which increases impedance (square of the number of turns)   4 passes will raise impedance to aprx. 275 ohms at 100 KHz & 2200 ohms at 2 MHz,   probably about as effective as it can be.  

FYI:  the other ferrite I use is N30 (MnZn) (the TDK ones)  N30 is a general purpose mix used in split clamp ferrites by most all brands in Europe & Asia, its working range is from 1 MHz to over 500 MHz peaking at 400 MHz,  too high for SMPS.   I use them on power, phone & Ethernet.   

John Swenson has some ideas about power supply ac>dc leakages and smps noise signature. He freely shares his findings and knowledge from the measurements he is trying to carry out. Curious minds can have a look at the computeraudiophile forums in the Uptone Audio sponsored threads e.g. - smps and grounding , the one on DIY DC cables and Sonore sponsored one - microRendu. He has a Tech Corner section on Uptone Audio website but I believe that has not updated that in a while.

Huge posted:

People have done this with Apple Airports.

I tried this several times but never got it to work...

Thanks Brilliant,  I have been reading his stuff;  I'm not going with his ground mod,  my iFi iPower on the Cisco switch is reputed to be the quietest of any PSU - SMPS & LPS - by some margin,  I fear adding an earth (ground) when its purposely been designed without it might screw up what it does & most important I could not hear any difference when I did try the mod.    

You are welcome - as you say, what you hear is what counts!

Huge posted:

What's often referred to as 'noise', is usually low-level non-harmonically related signal components.

It sounds like  something is riding on top of the high frequencies making them hotter e sounding, imparting a metallic twinge to them..

It's strange indeed - 256Kbps from spotify seems more relaxing to listen to than CD rips from the UnitiCore. Oh well - I guess I can take that as a positive and enjoy Spotify. More music on it anyway... :-)

It seems strange to me that such efforts will be gone too, to ruin ones home networking. If you don't like the sound from streaming stick with record decks, rather than the convoluted efforts to make things as complicated as possible to lower the 'noise floor'

garyi posted:

It seems strange to me that such efforts will be gone too, to ruin ones home networking. If you don't like the sound from streaming stick with record decks, rather than the convoluted efforts to make things as complicated as possible to lower the 'noise floor'

That'll spoil half the fun Garyi;    I do recall all the many mods & this weeks top tip of the record deck heydays, all manner of deck, arm, harness, cartridge & phono stage up grades,  magazines full of homebrew experiments some so bizarre it seemed we needed to be under the influence of either a higher spirit or mainlining something.

It's all about islation, I guess. I had a few more 'audiophile' ethernet cables around the house. I used them between the modem and the router, and between the router and the switch - all in the laundry room (that's where the internet comes into the house). 

That itself lowered the noisefloor a few notches..

Who know what gunk the Google WiFi will push into the ethernet cables.. so probably best not to go there..

At this point, the sound is still a little brighter than I'd like - and Spotify sounding more relaxing. 

Maybe replace the naca5s with super lumina. Lol!

Mango, I am intrigued what you mean with ‘isolation’ and what you ate trying to isolate from. When you say the noise floor is ‘lowered’, what noise floor are you referring to and how do you know it’s lowered. Subtle RF noise often has the effect of  dulling a sound or robbing it of high frequency info, perhaps a loss of timing info and clarity and making tiring to listen to. If you are finding a sound too bright or hot I suspect there are other issues at play.. perhaps grounding, system balance, or room reflections. Spotify vs local lossless streaming differences such that Spotify is more relaxing to listen does suggest to me more the latter... as Spotify will have less HF info when generally compared to local lossless streaming. 

It is probably is grounding issues of some sort - related to ethernet wiring in the walls - but I've got no idea on how to approach the issue. I figured I'll get around it by putting Google WiFi system in place - all wiring will be local.

Having said that, putting audiophile grade ethernet cables seem to add a layer of isolation - protecting the hifi gear from stuff flowing in from the local internet provider.

FWIW, this is the layout:

LaundryRoom (switching closet):
Modem ---AQ Vodka --> Router ---AQ Cinnamon --> Cisco switch. The cisco switch is in a router closet - and cables from it are plugged into the wall (this is un-powered).

Living Room:
NDS plugged straight into the wall using AQ Vodka.

Study:
272 plugged into the wall using Chord C-Stream.

Loft:
UnitiCore plugged into a Cisco switch which is then plugged into the wall.

So - the UnitiCore via SPDIF into the 272 doesn't have the artifacts that I get while playing upnp. Just to test this out, I brought the UnitiCore down in to the study. It'll probably stay here..

Makes sense to me Mango,because now you are wired directly to the 272,and the Ethernet to the Core is only used to give you remote,or iPad control.

This is the way mine is used,and I prefer it this way over UPNP.

It’s worth noting that different media servers can sound different with the current Naim streamers... I have found that inter frame timing variability leads to a subtle change  sonic character, akin to WAV vs FLAC. Clearly the frame timing on Ethernet is quite different from the frame timing on SPDIF, so I would expect subtle change in sonic character between the two.... i very much doubt your ISP is giving you physical network noise.. I suspect it’s more the other way around...

Mike-B posted:

Huge posted: 

To what breed of ferrets are you changing?

I'm changing some split clamps I had plastered around my SMPS's on both 230VAC & DC sides.   They originally came from my work life ...............  & I really did not know what the mix was, the doubt was a scratch that I had to itch.     

I've opted for Fair-Rite 75 as it's the lowest frequency ferrite mix in split clamp applications.  I have the SMPS DC wires wound around the ferrite cores which increases impedance (square of the number of turns)   4 passes will raise impedance to aprx. 275 ohms at 100 KHz & 2200 ohms at 2 MHz.  

@Huge, update re the above post,  I made the changes over a week ago & have been in/out away since so had not had a chance for serious listening.    Anyhow,  last evenings first real listening session streaming from NAS was all positive, all very subtle but a nice level of added smoothness & clarity,  very similar effect to what I heard when I installed the iFi iPower SMPS on the switch.  The iPower had the same new ferrite treatment but I'm not sure if its doing anything to an already almost silent SMPS,  I suspect the change is due to surrounding all my SMPS's (phone & broadband) on both AC & DC sides with ferrite 75.  Anyhow,  its made a difference.                        I've not changed the other ferrite N30 (the TDK's) I have on ethernet,  phone & power.  

Hi [@mention:1566878603907884],

With apologies to MM for the hijacking.

I tested my system by removing all the ferrets from the mains cables, then unplugging every SMPS I could find (and the fridge and freezer!) and then reconnecting them:  I couldn't detect any difference.  To check the network devices I compared them powered against unplugged using a USB stick.

There again my DC blocker also has a 250μH CM choke and a 10μH DM choke; so the ferrets I have on the power lines are probably rather redundant.  Also the Wurth Electronik 4 W 620 doped NiZn ferrets I use on the Ethernet cables still have a bit of effect some way below 1MHz - they're much better at MF than most NiZn ferrets.

Having said that the resolution of my system has increased now I've swapped the W2549 for the W2497 interconnects, so I should probably repeat the tests.  (But before that I need to replace my graphics card as my old FirePro V4900 is dying.)

Hi Huge,  I didn't mention that I had my grotty old 'scope on the AC & DC,  its a bit basic & hit & miss,  but the hash ripple on DC especially is reduced with the new ferrets,   Also worth mentioning all my SMPS's are powered off my UPS with its internal iso-transformer & C&M choke,  so the AC side ferrite is probably OTT,  but whatever, they're staying   ---------  end of thread highjack